Question

例如，我有以下代码：

def model_fn(features, labels, mode, params):
  """Model function for Estimator."""

  first_hidden_layer = tf.layers.dense(features["x"], 10, activation=tf.nn.relu)
  second_hidden_layer = tf.layers.dense(first_hidden_layer, 10, activation=tf.nn.relu)

  output_layer = tf.layers.dense(second_hidden_layer, 1)

  predictions = tf.reshape(output_layer, [-1])

  loss = tf.losses.mean_squared_error(labels, predictions)

  optimizer = tf.train.GradientDescentOptimizer(
      learning_rate=params["learning_rate"])
  train_op = optimizer.minimize(
      loss=loss, global_step=tf.train.get_global_step())

  return tf.estimator.EstimatorSpec(
      mode=mode,
      loss=loss,
      train_op=train_op,export_outputs=export_outputs)

train_input_fn = tf.estimator.inputs.numpy_input_fn(
  x={"x": np.array(training_set.data)},
  y=np.array(training_set.target),
  num_epochs=None,
  shuffle=True)

nn.train(input_fn=train_input_fn, steps=100)

如何获得＆＃34; second_hidden_layer＆＃34;的输出值，而不是张量而是实际值？我试图使用此代码但失败了。

export_outputs = {"en_out": tf.estimator.export.RegressionOutput( second_hidden_layer)}

Answer 1

tf.estimator.EstimatorSpec有另一个名为＆＃34;预测＆＃34;的参数。使用此dict可以直接调用Estimator返回预测。

predictions = {
    "en_out" : second_hidden_layer
}

并将其添加到EstimatorSpec中，如

return tf.estimator.EstimatorSpec(
    mode=mode,
    loss=loss,
    train_op=train_op,predictions=predictions)

参数＆＃34; export_outputs＆＃34;用于例如for tensorflow serving。

如何在tf.estimator.Estimator

1 个答案: